Panel : Cultural Heritage in the New Age of AI

Smart data means the trusted, contextualized, relevant, cognitive, predictive, and consumable data at any scale, revealing the most important V(value) of big data. A fundamental advancement of smart data that distinguishes it from other approaches is the methodology of big data, which assumes the ability to reveal the unknown-unknowns. The smart data concept is essential in relation to the role of libraries, archives, and museums (LAMs) in supporting Digital Humanities research.

In the new era of AI, smart data may be efficiently processed using AI technology while also being used by AI to generate intelligent applications and produce high-quality data. However, since information-bearing CH entities are usually non-textual and non-machine-readable, generating smart data for diverse CH resources presents complexities in data processing.

This introduction will bring some special cases in recent innovative research projects in the world, ranging from working on manuscripts, symbolic symbols, images, sound, smells, to museum specimens and other objects. Such advancement makes applications for promoting CH more abundant and practical.

This presentation reviews lessons learned on developing over 20 in-use systems in over 20 years for publishing and studying Cultural Heritage data on the Semantic Web, including museum collections, bibliographical data and manuscripts, military history, biographies and narratives, epistolary data, archaeology, law, parliamentary data, and art. Targeted to Digital Humanities researchers, application developers, and the public, these systems include a Linked Open Data (LOD) service with a live SPARQL endpoint that can be used with modest programming skills, and a ready-to-
use semantic portal on top of it. This work has evolved gradually into the so-called Sampo Model for creating and publishing Knowledge Graphs (KG) as LOD services and semantic portal User Interfaces (UI).

The scope of the digital archiving has drastically expanded to encompass various domains such as intangible cultural heritage, social events, natural and man-made disasters, and so on in the 21st century. This presentation will discuss data modeling to digitally curate and archive intangible entities in the cultural and historical domains. The One-to-One principle of Metadata is applied to the metadata modeling between the original objects (which may be intangible or ephemeral) / their surrogates and their metadata.

This presentation outlines the development of an ontology framework designed to enhance the organization and analysis of Qing Dynasty official document archives. The framework is built upon three key elements: First, the documents are treated as a distinct genre, where their form, function, and content are systematically analyzed to reveal their inherent structures. Second, scholars' research questions are transformed into competency questions (CQs), ensuring that the ontology meets the specific investigative needs of the domain. Third, the ontology is aligned with the CIDOC CRM standard, a widely adopted cultural heritage data model, to ensure interoperability and reusability within broader research frameworks. Together, these approaches enable the transformation of unstructured text into structured, machine-readable data, supporting advanced semantic retrieval and analysis in the digital humanities. Additionally, the potential application of Generative AI (GenAI) for entity recognition and data transformation will be discussed, illustrating how ontology-driven automation can further enhance research in this area.